Autonomous vehicles have received some bad press recently with some accidents and deaths that looked like the technology might be set back. Even some interested parties like Apple have gotten skittish about their development efforts. For example, earlier this month on September 7, Apple closed parts of its self-driving car project and laid off dozens of employees in that group. They are re-thinking their plan for self-driving vehicles. Then two weeks later on September 21, the day that US Federal regulators provided a 15-point checklist for self-driving cars, Apple announced that it had tripled its R&D budget to $10B and also announced possible plans to purchase McLaren, possibly beginning a rivalry to Elon Musk’s Tesla
This following US Department of Transportation checklist is actually a set of guidelines that will force tech companies and automobile manufacturers to prove that their offerings of semi-autonomous and fully-autonomous vehicles are safe and road-worthy before they ever reach the road surface. This is actually a technology driver for the industry since it is setting up of an official oversight that approves the operation of the autonomous vehicle on roads. Here is a brief summary of these 15 key points as broken down by the NY Times:
Data Sharing : Automobile manufacturers will need to share all the driving data stored in the car’s electronic memory with federal regulators after an accident of system failure. This will enable a reconstruction of what led to the accident or failure.
Privacy : Automobile manufacturers will need to notify buyer, in laymen’s terms, in advance of which data the vehicle’s computer will be storing. The buyer should retain the right to prevent the collection of personal biometric or driver behavioral information before purchasing the vehicle.
System Safety : The automobile’s electronic system will need to quickly respond to any software problems that may occur as well as to accident near-misses, traction problems or any other types of risk occurrences that may present themselves. Automakers may be held to bring in an outside expert party to examine and validate the safety system electronics and determine the system’s capability of remaining safe even in the event of any technological problem that may occur.
Digital Security : The vehicle electronic system must be able to prevent and deter online hacks of the system. Manufacturers will be required to record and share, with others in the industry,all programming testing results and programming decisions regarding security.
Human-Machine Interface (HMI) : The automobile must be proven to be able to safely switch from auto-pilot to driver control at any time. The driver should always be clearly advised when autonomous driving mode is not available. In addition, the vehicle will need to be able to somehow indicate to pedestrians and other vehicles as to when the autopilot is actively engaged. Finally, there is a stipulation that the autonomous system will need to be designed for people who are physically challenged.
Crashworthiness : The autonomous vehicle must be proven to meet the National Highway Traffic Safety Administration’s (NHTSA’s) standards for “crashworthiness”, that is, prove that the vehicle is best built to protect the vehicle occupants in the event of a crash.
Consumer Education : Automaker sales reps and staff must be trained as ti the workings of the autopilot so they can properly educate car dealers and distributors. Automakers and sellers will also need to supply consumers with training regarding the limitations and full capabilities of the autonomous vehicle as well as regarding emergency contingency scenarios.
Certification : All software updates and new driverless features must be submitted to the NHTSA.
Post-crash behavior : After the occurrence of an accident, automakers will need to prove the vehicle to be safe again. This will also help ensure that all damaged sensors or critical safety control systems damaged in the accident will be properly repaired, tested and restored to full functionality before being returned to the owner.
Laws and Practices : The vehicle will need to properly abide by state and local laws and practices that apply to the driver. Recognizing local speed limits in different cities and states, whether U-turns are lawful or right turn on red is lawful in a region or state. Also, the vehicle will need to be able to respond to avoiding a crash by possibly violating a law like crossing over a double-yellow line. This also brings up the question whether an autonomous vehicle will put other vehicles in danger to save itself. A very controversial question. See the next paragraph.
Ethical Considerations : When a person is driving a car, there are driving decisions to be made that may have ethical considerations. The method in which an autonomous vehicle system is programmed will also carry ethical consequences. As I mentioned in the previous paragraph, should an autonomous vehicle protect its occupants at the expense of other cars and their occupants on, in or around the road? Or while in heavy traffic, should a car be allowed to violate the crossing of a double line law if there is risk of oncoming traffic? These types of decisions need to be disclosed to the NHTSA before the car is approved for the road.
Operational Design : This is a type of manual like the printed manuals in a standard car. The manufacturer must where, when and under which conditions an autonomous system functions. Manufacturers will need to prove that the vehicle has been thoroughly tested to and validated to meet these conditions such as how fast a car can travel and if it is able to drive at night and on a rocky dirt road.
Detection and Response : This point discusses how the vehicle will respond to other cars, pedestrians, animals, and things like falling trees or rocks. The manufacturer will need to demonstrate that the car has been programmed to respond to normal driving conditions such as lane changing and minding traffic signals. The automaker must also prove that the vehicle will be able to avoid any unforeseen events and crashes.
Fallback : The autonomous vehicle should be able to change modes safely when a technical malfunction occurs. The change from autonomous to human control must be able to take into account the driver’s condition such as being under the influence of alcohol or drugs or being drowsy and not able to take over control of the car safely.
Validation : The auto manufacturers will be held responsible to develop testing and validation methods that will account for the wide range of technologies in an autonomous vehicle. The tests should include at a minimum, simulation, operation on a test track, and road testing.
In my opinion, these mandates show a firm government interest in increasing highway safety via the use of autonomous vehicles. This will surely help boost the industry development of safe, fully functioning autonomous vehicle systems. Ultimately they will be safer than human, distracted drivers and lower accident rates.
What are your thoughts on these mandates. Are they enough or too restricting?